Motor operation energy efficiency detection method and device

Abstract

The invention provides a motor operation energy efficiency detection method and device, and belongs to the technical field of motor detection, and the method comprises the steps: determining a plurality of normal samples and a plurality of abnormal samples of motor operation parameters based on a clustering algorithm; performing sample expansion on the plurality of abnormal samples according to the sample features of the plurality of abnormal samples to obtain a plurality of expanded abnormal samples; and training a state detection model based on the plurality of normal samples and the plurality of expanded abnormal samples, and detecting the running state of the to-be-detected motor according to the trained state detection model. According to the motor operation energy efficiency detection method and device provided by the invention, the accuracy of motor operation energy efficiency detection can be improved.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of motor detection, and more specifically, to a method and device for detecting the energy efficiency of a motor operation. Background Art

[0002] In modern industrial production and daily life, motors are widely used in various scenarios as key equipment for converting electrical energy into mechanical energy. From the operation of large mechanical equipment in factories to the normal operation of household appliances, motors are everywhere. The operating status of motors directly affects the overall performance, production efficiency and operational safety of the equipment.

[0003] Considering that the abnormal detection of motor operation depends on a large number of abnormal samples, but in practical applications the sample size of abnormal samples is insufficient, the existing motor operation energy efficiency detection method has the problem of low accuracy, which affects the stability of motor operation. Summary of the invention

[0004] The purpose of the present disclosure is to provide a method and device for detecting the energy efficiency of a motor operation, so as to improve the accuracy of the detection of the energy efficiency of a motor operation.

[0005] According to a first aspect of an embodiment of the present disclosure, a method for detecting energy efficiency of a motor is provided, comprising: Determine multiple normal samples and multiple abnormal samples of motor operation parameters based on a clustering algorithm; Performing sample expansion on the multiple abnormal samples according to the sample features of the multiple abnormal samples to obtain multiple expanded abnormal samples; The state detection model is trained based on the multiple normal samples and the multiple expanded abnormal samples, and the motor operation energy efficiency is determined by detecting the motor operation state to be detected according to the trained state detection model.

[0006] According to a second aspect of the embodiments of the present disclosure, a motor operation energy efficiency detection device is provided, comprising: A sample determination module, used for determining a plurality of normal samples and a plurality of abnormal samples of motor operation parameters based on a clustering algorithm; A sample expansion module, used to expand the plurality of abnormal samples according to the sample characteristics of the plurality of abnormal samples to obtain a plurality of expanded abnormal samples; The detection model building module is used to train the state detection model based on the multiple normal samples and the multiple expanded abnormal samples, and to detect the motor operation state to be detected according to the trained state detection model to determine the motor operation energy efficiency.

[0007] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the above-mentioned motor operation energy efficiency detection method when executing the computer program.

[0008] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned motor operation energy efficiency detection method are implemented.

[0009] The beneficial effects of the motor operation energy efficiency detection method and device provided by the embodiments of the present disclosure are: The present disclosure screens samples based on a clustering algorithm, can accurately distinguish normal and abnormal samples, classify motor operating parameters according to characteristics, and provide a reliable data basis for subsequent state detection. Compared with traditional manual judgment, it is more objective and accurate, and effectively avoids the interference of human factors. At the same time, the present disclosure greatly enriches the diversity and quantity of abnormal data by expanding abnormal samples. In practice, abnormal samples are relatively scarce. After expansion, the state detection model can learn more potential fault modes, effectively improving the adaptability and fault diagnosis ability of the model. Finally, the present disclosure uses normal samples and expanded abnormal samples to train the model, so that the model can learn more comprehensive and accurate motor operating state characteristics. When the model training is completed, when the operating state of the motor to be detected is detected, the operating state of the motor to be detected can be more accurately detected. Based on the operating state of the motor, the change of the motor operating energy efficiency can be accurately judged. Therefore, the present disclosure can improve the accuracy of motor operating energy efficiency detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flow chart of a method for detecting motor operation energy efficiency provided by an embodiment of the present disclosure; Figure 2 A structural block diagram of a motor operation energy efficiency detection device provided by an embodiment of the present disclosure; Figure 3 A schematic block diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0012] In the following description, specific details such as specific system structures and technologies are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present disclosure. However, it should be clear to those skilled in the art that the present disclosure may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obstructing the description of the present disclosure with unnecessary details.

[0013] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0014] Please refer to Figure 1 , Figure 1 A flow chart of a method for detecting motor energy efficiency according to an embodiment of the present disclosure is provided, the method comprising: S101: Determine a plurality of normal samples and a plurality of abnormal samples of motor operating parameters based on a clustering algorithm.

[0015] In this embodiment, the clustering algorithm is an unsupervised learning algorithm for dividing the data points in the data set into different groups or clusters according to similarity. The clustering algorithm can group the data points according to the similarity of the characteristics of the motor operating parameters. For example, the K-Means clustering algorithm can determine the cluster to which each data point belongs by calculating the distance between the data point and each cluster center.

[0016] Motor operating parameters are various quantitative values ​​that can reflect the motor operating status, including but not limited to the motor's temperature, noise and vibration, as well as line voltage, current, power metering and power quality parameters. Changes in the above parameters can help determine whether the motor is operating normally.

[0017] The normal sample is a data sample composed of the operating parameters corresponding to the motor under normal operating conditions. The motor operating parameters of the normal sample should meet the standard range of normal motor operation. For example, during normal operation, the motor current fluctuates within a certain range of the rated current, and the temperature is also within the normal operating temperature range.

[0018] Abnormal samples are data samples composed of the corresponding operating parameters of the motor when it fails or does not operate normally. The parameter values ​​of the operating parameters of the abnormal samples may exceed the normal standard range, or the relationship between the operating parameters does not conform to the rules of normal operation. For example, when the motor winding is short-circuited, the current will suddenly increase. The corresponding abnormal samples can record this abnormal current value and changes in other related parameters.

[0019] Specifically, in this embodiment, the motor operating parameter data over a period of time is first collected. The above data can be acquired in real time by various sensors installed on the motor (such as current sensors, temperature sensors, acceleration sensors, etc.); then, the above data is used as input and processed by a clustering algorithm. The clustering algorithm will divide the above motor operating parameters into different clusters according to the similarity between the data.

[0020] S102: Expand the multiple abnormal samples according to the sample features of the multiple abnormal samples to obtain multiple expanded abnormal samples.

[0021] In this embodiment, the sample characteristics of the abnormal sample are various characteristics contained in the abnormal sample that can reflect the abnormal operating state of the motor. The above characteristics can be special values ​​of a single motor operating parameter, abnormal relationships between motor operating parameters, or changing trends of motor operating parameters over time, etc.

[0022] Sample expansion is to generate more samples with similar characteristics to the above abnormal samples based on the existing real abnormal samples through data expansion methods, so as to increase the number and diversity of abnormal samples, thereby providing richer data for subsequent model training. The data expansion method used in this embodiment can be a generative adversarial network.

[0023] This embodiment first analyzes multiple existing abnormal samples to find their common or representative sample features. Then, this embodiment uses the above sample features and adopts a suitable data expansion method to generate new samples. The newly generated samples have similar features to the existing samples, and the expanded abnormal samples are finally obtained through the collection.

[0024] Assume that this embodiment has determined a group of samples of abnormal motor operation, which are about motor bearing failure. The sample characteristics are mainly reflected in the motor vibration acceleration parameters. The normal motor vibration acceleration amplitude is between 0-5m / s², while the vibration acceleration amplitude of these abnormal samples is between 10-20m / s², and the vibration acceleration frequency has a peak value between 100-200Hz.

[0025] In order to expand the sample, this embodiment can use data interpolation, random noise addition and other methods. For example, for the vibration acceleration amplitude, a new value is randomly generated within the range of 10-20m / s²; for the frequency peak, a new frequency peak position is generated between 100-200Hz according to a certain probability distribution (such as normal distribution).

[0026] Assuming there are 10 original abnormal samples, through the above method, each sample generates 3 new samples, thus obtaining 30 expanded abnormal samples. These expanded samples also reflect the abnormal characteristics of vibration acceleration amplitude and frequency peak value when the motor bearing fails, and can provide richer data for subsequent state detection model training.

[0027] S103: training a state detection model based on a plurality of normal samples and a plurality of expanded abnormal samples, and detecting the motor operation state to be detected according to the trained state detection model to determine the motor operation energy efficiency.

[0028] In this embodiment, the input set is a plurality of normal samples and a plurality of expanded abnormal samples, and the output is the corresponding motor operating state. The state detection model is a model for identifying the motor operating state, which can be constructed based on machine learning or deep learning, such as a support vector machine, a neural network, etc. The state detection model is to establish a judgment criterion for distinguishing between normal operating state and abnormal operating state by learning the characteristics of normal samples and abnormal samples. The trained state detection model is a state detection model that has been trained, and after testing and verification evaluation, the accuracy rate is greater than or equal to the preset threshold.

[0029] The motor operating state to be detected is the current state of the motor that needs to be judged in actual applications. The operating parameters of the motor are collected in real time through multiple sensors, and then the above parameters are input into the state detection model to determine whether the motor is in a normal operating state.

[0030] A large number of normal samples and expanded abnormal samples collected are used as training data and input into the state detection model. During the training process, the state detection model will adjust its own parameters (such as weights and biases in the neural network) according to the characteristics of the above samples, and continuously learn the differences between normal samples and abnormal samples, thereby building a judgment model that can accurately distinguish between normal and abnormal operating states.

[0031] When the state detection model reaches a certain accuracy and stability after training, the trained state detection model is obtained. For the motor in actual operation, its operating parameters are obtained in real time through multiple sensors, and these parameters are input into the trained state detection model. The state detection model will output a judgment result based on the previously learned feature pattern, telling whether the current operating state of the motor is normal or abnormal.

[0032] Training phase: Assume that this embodiment uses a multilayer perceptron (MLP) as a state detection model. 1,000 normal samples were collected, in which the motor current was between 95% and 105% of the rated current, the voltage was between 98% and 102% of the rated voltage, and the speed was stable within ±3% of the rated speed. At the same time, there are 500 abnormal samples obtained after expansion, covering the operating parameter characteristics of different fault types such as motor overload (such as current exceeding 120% of the rated current), winding short circuit (such as a sharp increase in current and unbalanced three-phase current), and bearing fault (such as a substantial increase in vibration value).

[0033] These normal samples and abnormal samples are divided into training set and validation set according to a certain ratio (such as 80% for training and 20% for validation). The training set is input into the MLP model, and the number of nodes in the input layer of the model is set to be the same as the number of motor operating parameters (assuming 5 parameters, such as current, voltage, speed, temperature, and vibration). The hidden layer is set to 2 layers, with 10 and 8 nodes in each layer, and 1 node in the output layer (indicating normal or abnormal judgment results, 0 represents normal, and 1 represents abnormal). During the training process, the stochastic gradient descent algorithm is used to continuously adjust the weights and biases of the model, so that the prediction accuracy of the model on the training set is continuously improved. After multiple iterations of training, when the accuracy of the model on the validation set reaches more than 95%, the model training is considered complete.

[0034] Detection phase: For a running motor, multiple sensors installed on the motor collect its current, voltage, speed, temperature, vibration and other parameters in real time. Assume that the collected parameters are: the current is 110% of the rated current, the voltage is 99% of the rated voltage, the speed is 98% of the rated speed, the temperature is 10°C higher than the normal operating temperature, and the vibration value exceeds the normal range by 2 times. These parameters are input into the trained MLP model.

[0035] After calculation, the model outputs a result of 1, indicating that the motor is currently in an abnormal operating state. Further analysis shows that the abnormal characteristics of its current, temperature, and vibration values ​​are consistent with the abnormal sample characteristics of overload and bearing failure in the training set, so it can be judged that the motor may have overload and bearing failure problems, and timely maintenance and repair are required.

[0036] Therefore, the motor operation state is detected according to the trained state detection model to obtain the detection result of the motor operation state; If the detection result of the motor operation state is normal operation, the motor operation energy efficiency is stable; If the detection result of the motor operation state is abnormal operation, the fault type of the motor operation is determined, and based on the fault type, the motor operation energy efficiency is determined to be reduced.

[0037] When the motor operates normally, the motor's operating energy efficiency is better and is in a relatively stable state, and can operate efficiently; when the motor operates abnormally, the motor's operating energy efficiency decreases and the loss increases.

[0038] It can be concluded from the above that the present disclosure can accurately distinguish normal and abnormal samples by screening samples based on clustering algorithms, classify motor operating parameters according to characteristics, and provide a reliable data basis for subsequent state detection. Compared with traditional manual judgment, it is more objective and accurate, and effectively avoids the interference of human factors. At the same time, the present disclosure greatly enriches the diversity and quantity of abnormal data by expanding abnormal samples. In practice, abnormal samples are relatively scarce. After expansion, the state detection model can learn more potential fault modes, effectively improving the adaptability and fault diagnosis ability of the model. Finally, the present disclosure uses normal samples and expanded abnormal samples to train the model, so that the model can learn more comprehensive and accurate motor operating state characteristics. When the model training is completed, when the operating state of the motor to be detected is detected, the operating state of the motor to be detected can be more accurately detected. Based on the operating state of the motor, the change of the motor operating energy efficiency can be accurately judged. Therefore, the present disclosure can improve the accuracy of motor operating energy efficiency detection.

[0039] In one embodiment of the present disclosure, the motor operating parameters include multiple types of operating parameters; Based on the clustering algorithm, multiple normal samples and multiple abnormal samples of motor operating parameters are determined, including: Perform clustering processing based on the parameter characteristics of each type of operating parameters to obtain clustering results corresponding to each type of operating parameters, and the clustering results include a normal sample set and an abnormal sample set; The normal sample sets corresponding to each type of operating parameters are collected to obtain multiple normal samples; The abnormal sample sets corresponding to each type of operating parameters are collected to obtain multiple abnormal samples.

[0040] In this embodiment, the multiple types of operating parameters are motor operating parameters that include parameters of multiple different properties, not a single type, and each type of parameter has its own characteristics and changing rules. The parameter characteristics are the characteristics and properties of each type of operating parameter, including the value range, changing trend, and fluctuation law of the parameter. For example, the parameter characteristics of the current may fluctuate in a small range around the rated current during normal operation, and may rise or fall sharply during a fault.

[0041] This embodiment uses a clustering algorithm to perform grouping operations on each type of operating parameters. Each type of operating parameters includes normal operating parameters and abnormal operating parameters. The clustering algorithm can divide each type of operating parameters into different clusters according to the similarity between normal operating parameters or abnormal operating parameters, so that the data in the same cluster has a high similarity, while the data between different clusters are relatively different.

[0042] The normal sample set corresponding to each type of operating parameter is a set of samples of the normal operating state of the motor. In this sample set, the corresponding motor operating parameters are all within the normal range, which meets the characteristics of the normal operation of the motor.

[0043] The abnormal sample set corresponding to each type of operating parameter is a sample set when the motor is in an abnormal operating state. The motor operating parameters corresponding to the sample set deviate from the normal range and show abnormal or fault characteristics.

[0044] Specifically, the steps of this embodiment are as follows: First, for each type of motor operating parameter (such as current, voltage, speed, etc.), analyze its parameter characteristics separately, and then use the clustering algorithm to process it. After clustering, the data of each type of operating parameter will be divided into different groups, one of which represents the sample set of the normal operating state of the motor, that is, the normal sample set; the other group represents the sample set of the abnormal operating state of the motor, that is, the abnormal sample set. For example, for the temperature parameters of the motor, the clustering algorithm will classify the data within the normal temperature range as the normal sample set according to the change of the temperature value, and classify the data that exceeds the normal range and shows abnormal increase or decrease as the abnormal sample set.

[0045] Second, the normal sample sets obtained after clustering each type of operating parameter are combined together. Because each type of normal sample set only contains normal samples of that type of parameter, through set operations, all types of normal samples can be integrated to form a sample set containing normal state information of multiple motor operating parameters. This set is multiple normal samples. For example, the normal current sample set, the normal voltage sample set, the normal speed sample set, etc. are combined to obtain a sample set that comprehensively reflects the state of each parameter when the motor is operating normally.

[0046] Third, by merging the abnormal sample sets obtained after clustering each type of operating parameter, a sample set containing abnormal state information of various motor operating parameters, i.e., multiple abnormal samples, can be obtained. For example, the current abnormal sample set, temperature abnormal sample set, speed abnormal sample set, etc. can be merged to form a sample set that comprehensively reflects various abnormal operating states of the motor.

[0047] Assume that the motor operating parameters include current, voltage and speed.

[0048] For the current parameter, the current data was collected over a period of time, and its parameter characteristics showed that it fluctuated between 95% and 105% of the rated current during normal operation. The hierarchical clustering algorithm was used to cluster these current data, and the data within the range of 95%-105% of the rated current was classified as a normal sample set, while the data outside this range was classified as an abnormal sample set.

[0049] For voltage parameters, the voltage is between 98% and 102% of the rated voltage during normal operation. The hierarchical clustering algorithm is also used to classify the voltage data within this range as a normal sample set, and the others as an abnormal sample set.

[0050] For the speed parameter, the speed is stable within ±3% of the rated speed during normal operation. Through clustering, the speed data that meets this range is classified as a normal sample set, and those that exceed the range are classified as an abnormal sample set.

[0051] After that, the normal sample sets of current, voltage, and speed are combined to obtain multiple normal samples, which fully reflect the current, voltage, and speed status of the motor during normal operation. The abnormal sample sets of current, voltage, and speed are combined to obtain multiple abnormal samples, which cover various situations when the motor has abnormal current, voltage, and speed. For example, the abnormal samples may contain different types of abnormal status information such as excessive current, low voltage, and unstable speed.

[0052] It can be concluded from the above that this embodiment can systematically analyze multiple types of motor operating parameters and accurately distinguish normal and abnormal motor operating states through clustering processing. This embodiment not only improves the accuracy and efficiency of fault detection, but also provides reliable data support for subsequent motor maintenance and management. In addition, this method avoids the limitations of traditional methods that rely on manual experience judgment, and realizes automated and intelligent detection of motor operating states, which is of great significance for ensuring stable operation of the motor and preventing potential faults.

[0053] In one embodiment of the present disclosure, clustering processing is performed based on parameter features of each type of operating parameters to obtain clustering results corresponding to each type of operating parameters, including: Determine multiple cluster centers; Determine the distance between each type of operating parameter and the operating parameters of multiple cluster centers based on the first distance measurement method to obtain multiple first distances; A clustering result corresponding to each type of operating parameter is determined based on the multiple first distances.

[0054] In this embodiment, the cluster center is a representative point of each cluster. In the data space, the cluster center can be regarded as a data point, and its coordinate value represents the average feature or typical feature of the cluster data.

[0055] The first distance measurement method is used to measure the calculation formula of the difference between two data points. In this embodiment, the first distance measurement method is used to measure the difference between each type of operating parameters of the motor and the operating parameters of the cluster center.

[0056] Assume the motor has Class operation parameters are For each type of operating parameters , the value in a certain period of time is , is the number of samples, and the corresponding parameter value of the cluster center is .

[0057] First, the sum of squared Euclidean distances between each type of operating parameter and the cluster center is calculated. Class operating parameters, the calculation formula of Euclidean distance is:

[0058] in, Indicates Class operating parameters The sample value of Indicates Class operating parameters The value of the cluster center.

[0059] Considering the different importance of different types of operating parameters to the motor operating status judgment, a weight is set for each type of operating parameter. , The weight setting can be determined based on the actual operation of the motor and experience. For example, the current parameter is more sensitive to motor failure, and its weight can be set relatively high.

[0060] At the same time, considering the correlation between motor operating parameters, the correlation coefficient adjustment term is introduced. The average correlation coefficient of the class operating parameters with all other class operating parameters , and the average correlation coefficient of the corresponding parameters of the cluster center . Correlation coefficient The calculation formula is:

[0061] in, Indicates Motor operating parameters of the class, Indicates Motor operating parameters of the class , represents the number of samples, Indicates Class operating parameters Sample values, Indicates Class operating parameters Sample values, Indicates The mean of the sample values ​​of the class running parameter, Indicates The mean of the sample values ​​of the class run parameter.

[0062] The calculation formula of the first distance metric as follows:

[0063] in, represents the adjustment coefficient, , which is used to measure the impact of the Euclidean distance and correlation coefficient adjustment terms. When it is close to 1, it focuses more on the influence of Euclidean distance; when When it is close to 0, the correlation coefficient adjustment term has a greater effect.

[0064] In this embodiment, the weight is set , the calculation formula of the first distance metric can differentiate the importance of judging the motor operating status according to different operating parameters. For example, when judging whether the motor is overloaded, the current parameter is more critical than the temperature parameter. Therefore, a larger weight can be set for the current parameter so that the first distance metric can better reflect the characteristics related to overload. The correlation coefficient adjustment item is introduced to take into account that the motor operating parameters do not exist in isolation, but are interrelated. For example, when the motor load increases, the current and speed usually show a certain correlation change. By comparing the correlation coefficient difference between the operating parameters and the corresponding parameters of the cluster center, the similarity between the sample and the cluster center can be more comprehensively measured, thereby improving the accuracy of clustering. Adjustment coefficient The existence of can flexibly adjust the relative importance of Euclidean distance and correlation coefficient adjustment items in distance measurement according to actual conditions. In the case of strong correlation between motor operating parameters, it can be appropriately reduced When the parameter correlation is weak, increase To highlight the influence of Euclidean distance.

[0065] For example, weight The setting of is based on the prior knowledge of the importance of different operating parameters. For example, the importance of current parameters for judging the motor operating status is greater than that of temperature parameters, and the importance of temperature parameters for judging the motor operating status is greater than that of speed parameters, so the weight of current parameters is set to , the weight of the temperature parameter is , the weight of the speed parameter is The above weight distribution enables the first distance metric to more reasonably reflect the contribution of each parameter to the overall distance. If the current parameter has a large deviation, it will have a greater impact on the distance calculation due to its large weight, which is consistent with the fact that the current parameter is more critical to the motor operation status judgment in actual situations.

[0066] The sum of the squares of the Euclidean distances between each type of operating parameter and the cluster center measures the degree of dispersion of the sample points from the cluster center in each dimension of the motor operating parameters. Taking the current parameter as an example, the smaller the sum of the squares of the Euclidean distances, the closer the current parameter values ​​of the sample are to the current values ​​of the cluster center as a whole, that is, the higher the similarity between the sample and the cluster center in the current dimension. The Euclidean distance itself has good mathematical properties and is an intuitive distance measurement method that can accurately reflect the geometric distance between two points in space. Among the multi-dimensional parameters of motor operating parameters such as current, temperature, speed, etc., the Euclidean distance can comprehensively consider the differences in various dimensions, thereby providing a comprehensive measurement standard for the distance between the sample and the cluster center.

[0067] The correlation coefficient measures the linear correlation between different operating parameters. During the operation of the motor, there is often a certain correlation between different parameters. For example, current and temperature may have a certain positive correlation due to factors such as the internal resistance heating of the motor. By calculating the average correlation coefficient and the average correlation coefficient of the parameters corresponding to the cluster center, the influence of the mutual relationship between the parameters on the distance metric can be considered. When the correlation between the parameters of the sample is significantly different from the correlation between the parameters of the cluster center, the correlation coefficient adjustment item will increase the first distance metric accordingly. For example, if the correlation between the current and temperature of a sample is significantly different from the correlation between the current and temperature of the cluster center, it indicates that the sample has a large difference in operation mode from the typical mode represented by the cluster center. The correlation coefficient adjustment item can capture this difference more keenly, thereby improving the first distance metric's ability to identify sample abnormalities or differences. Therefore, the combination of Euclidean distance and correlation coefficient can improve the reliability and accuracy of motor operation status judgment.

[0068] The adjustment coefficient balances the weight of the weighted Euclidean distance and the correlation coefficient adjustment item in the entire distance metric. By adjusting the value of the adjustment coefficient, the performance of the first distance metric formula can be optimized according to the data characteristics of the motor operating parameters and the actual operating scenario of the motor. For example, when the distribution of the motor operating parameter data is mainly affected by the discrete degree of the parameter value itself, and the correlation between the motor operating parameters is relatively stable, the importance of the Euclidean distance part in the first distance metric can be appropriately increased; conversely, when the correlation change between the motor operating parameters is more critical to the abnormal detection of each sample point, the role of the correlation coefficient adjustment item should be highlighted. The above-mentioned dynamic adjustment of the adjustment coefficient based on the actual situation of the motor operation can adapt to different application scenarios and improve the reliability of the judgment of the motor operation status.

[0069] In this embodiment, the distance between each type of operating parameter and the operating parameters of multiple cluster centers is calculated by the first distance measurement method, and the obtained series of distance values ​​are multiple first distances. The clustering result is the result of dividing the data points into different clusters after being processed by the clustering algorithm. In the clustering of the motor operating parameters, the clustering result will indicate which operating parameters belong to the normal sample set and which belong to the abnormal sample set.

[0070] It can be concluded from the above that the determination of multiple cluster centers in this embodiment provides a benchmark for subsequent distance calculations, ensuring the diversity and representativeness of clustering. This embodiment uses the first distance measurement method to accurately calculate the distance between each type of operating parameter and the cluster center, so that the clustering results are closer to the actual data distribution. Ultimately, the clustering results determined based on the first distance not only help to identify the similarities and differences between parameters, but also provide strong support for subsequent pattern recognition, fault diagnosis and other applications, greatly enhancing data-driven decision-making capabilities.

[0071] In one embodiment of the present disclosure, sample expansion is performed on multiple abnormal samples according to sample features of the multiple abnormal samples to obtain multiple expanded abnormal samples, including: Input multiple abnormal samples into the target generative adversarial network model to obtain multiple expanded abnormal samples; Among them, the target generative adversarial network model learns the sample features of multiple abnormal samples.

[0072] In this embodiment, the target Generative Adversarial Networks (GAN) model is a trained GAN model. In this embodiment, the target GAN model generates new samples (i.e., expanded abnormal samples) based on existing abnormal samples. The new samples have similar characteristics to the original abnormal samples, thereby increasing the number and diversity of abnormal samples, which is helpful for subsequent more comprehensive training of state detection models and other applications.

[0073] In this embodiment, the collected sample data reflecting the abnormal operation state of the motor are provided as input to the constructed target GAN model. The model will generate new data with similar characteristics to the original abnormal samples based on the characteristic rules of these abnormal samples learned by itself. These newly generated data are the expanded abnormal samples.

[0074] During the training process of the GAN model, the generator will try to capture various features presented by multiple abnormal samples, such as the numerical range of the data, the trend of change, the relationship between different parameters, etc. The discriminator helps guide the generator to better learn these features. By constantly judging the difference between the generated data and the real abnormal samples, the generator is prompted to make continuous adjustments, and finally the generator can generate data that meets the characteristics of the abnormal samples, so as to achieve the purpose of learning and mastering these features.

[0075] It can be concluded from the above that this embodiment can accurately capture and reproduce the complex features of abnormal samples through the learning ability of the GAN model, thereby generating high-quality expanded samples. It not only greatly enriches the abnormal sample data set and enhances the generalization ability of the model, but also effectively alleviates the training difficulties caused by the scarcity of abnormal samples. In addition, the expanded abnormal sample set provides a more solid data foundation for subsequent tasks such as anomaly detection and classification, improves the performance and accuracy of the overall system, and has important practical application value.

[0076] In one embodiment of the present disclosure, a generative adversarial network model includes a generator and a discriminator; The motor operation energy efficiency detection method further includes: The generator is trained based on the historical abnormal sample set to obtain a trained generator; Inputting the random noise vector into the trained generator to obtain the first abnormal data set; Training the discriminator based on the historical abnormal sample set and the first abnormal data set to obtain a trained discriminator; Determine the target generative adversarial network model based on the trained generator and the trained discriminator.

[0077] In one embodiment of the present disclosure, the motor operation energy efficiency detection method further includes: Determining a confidence level for the first anomaly data set based on the discriminator; If the confidence is greater than the first threshold, the parameters of the generator are retained; If the confidence is less than or equal to the first threshold, the parameters of the generator are updated.

[0078] In this embodiment, the GAN model is a deep learning model architecture, which consists of two parts: a generator and a discriminator. The generator aims to generate data that is similar to the distribution of real data, and the discriminator is responsible for distinguishing whether the input data comes from the real data set or is generated by the generator. The two are trained in a way of mutual confrontation and continuous game, so that the data generated by the generator is closer and closer to the real data.

[0079] The generator receives a random noise vector as input, and outputs data that simulates the characteristics of a historical anomaly sample set through its internal neural network structure and parameter transformation. The discriminator takes as input a historical anomaly sample set (i.e., real data) or a simulated historical anomaly sample set generated by the generator. After being processed by its own network structure, it outputs a probability value, indicating the likelihood that the input data is real data.

[0080] The historical abnormal sample set is a data set of operating parameter data collected in the past that can represent the abnormal operating state of the motor. These samples contain parameter conditions corresponding to various fault types, such as the values ​​of current, voltage, temperature, vibration and other parameters under different faults. They are an important basis for training the GAN model.

[0081] The random noise vector is a set of numerical vectors randomly sampled from a specific distribution (such as normal distribution, uniform distribution, etc.). It is used as the input of the generator to provide a random basis for the generator to generate different data. The first abnormal data set is a parameter data set that simulates the abnormal operation state of the motor and is generated by the trained generator based on the input random noise vector. It is a data set generated by the generator based on the learned historical abnormal sample features.

[0082] The confidence is the probability value that the discriminator determines that the data in the first abnormal data set is real data, which reflects the degree of recognition of the discriminator on the data generated by the generator. The higher the probability, the closer the discriminator believes the data is to the real abnormal sample data.

[0083] The first threshold is a pre-set probability critical value, which is used to judge whether the quality of the data generated by the generator meets the requirements. When the confidence given by the discriminator is higher than this threshold, it is considered that the data generated by the generator meets the requirements; otherwise, it is considered that the data generated by the generator needs to be improved.

[0084] Specifically, the steps of this embodiment are as follows: First, provide the historical abnormal sample set (i.e., motor operation abnormal parameter data) to the generator, so that the generator can learn the characteristics, distribution patterns and other information of these data. By continuously adjusting its own neural network parameters (such as weights, biases, etc.), the generator can generate data similar to the characteristics of historical abnormal samples based on the input random noise vector. After a certain round of training, the result is a trained generator.

[0085] Second, after the generator is trained, multiple random noise vectors are randomly extracted from the set distribution (such as normal distribution), and then these vectors are input into the trained generator one by one. The generator processes each input random noise vector according to the characteristics of historical abnormal samples learned previously, and outputs the corresponding parameter data of the simulated abnormal operating state of the motor. These output data sets together constitute the first abnormal data set.

[0086] Third, the historical anomaly sample set (i.e., real anomaly sample data) and the data in the first anomaly data set generated by the generator are mixed together and then input into the discriminator. The discriminator learns based on these data and adjusts its own parameters in order to accurately distinguish which are real anomaly sample data and which are data generated by the generator. After a certain training process, a trained discriminator that can make better judgments is obtained.

[0087] Fourth, after the generator and discriminator are trained separately, they are combined to form the target GAN model. This model has the ability to generate simulated motor abnormal operation status data and determine the authenticity of the data. The target GAN model can be used to expand the motor abnormal operation samples and other operations.

[0088] Input the data in the first anomaly data set into the trained discriminator one by one. The discriminator will output a probability value for each input data. This probability value is the confidence of the discriminator that the data is real anomaly sample data, reflecting the discriminator's judgment on the quality of the data generated by the generator.

[0089] The confidence calculation formula is:

[0090] in, Indicates that the last layer of the discriminator neural network is input sample The linear combination output of , represents the weight vector, represents the input data vector, represents the bias, Indicates transpose.

[0091] A first threshold is set in advance, for example, 0.9. If the confidence output by the discriminator for a certain data in the first anomaly data set is greater than this threshold, it means that the data generated by the generator is of relatively high quality and closer to the real anomaly sample data. At this time, the existing parameters of the generator are temporarily kept unchanged because the performance of the current generator meets the expected requirements.

[0092] If the confidence level of the discriminator's output for a certain data in the first abnormal data set is less than or equal to the preset first threshold, it means that the data generated by the generator is still different from the actual abnormal sample data and the quality is not ideal. Therefore, it is necessary to update and adjust the parameters of the generator, such as by using the back-propagation algorithm to change the generator's neural network weights, bias and other parameters according to the feedback from the discriminator, in the hope that the generator can generate better quality data in the future.

[0093] From the above, it can be concluded that this embodiment trains the generator based on the historical abnormal sample set and generates the first abnormal data set in combination with the random noise vector. This method effectively simulates the abnormal state in the operation of the motor and enhances the generalization ability of the model. It not only improves the detection sensitivity of unknown abnormal situations, but also reduces the false alarm rate. Secondly, the discriminator is trained in combination with the historical abnormal sample set and the generated first abnormal data set, so that the discriminator can more accurately identify real abnormalities and simulated abnormalities, thereby improving the accuracy of motor operation energy efficiency detection. In addition, this embodiment introduces a confidence judgment mechanism to dynamically adjust the parameters of the generator, thereby ensuring the continuous optimization of the GAN model, and the adaptive adjustment strategy further improves the stability and reliability of the model.

[0094] In one embodiment of the present disclosure, the motor operation energy efficiency detection method further includes: Based on the result of the state detection, the motor operation state is warned to obtain the warning result; Remotely report warning results.

[0095] In this embodiment, the state detection result is a judgment conclusion obtained after analyzing the motor operating parameters through the state detection model, indicating whether the motor is currently in a normal operating state or there is an abnormal situation.

[0096] The motor operation status warning is based on the status detection results. When the motor operation status is found to be abnormal, the corresponding warning mechanism is activated to send an alarm message to the relevant personnel or system to remind them to pay attention to the operation status of the motor so that timely measures can be taken. The warning can be achieved in many ways, such as sound prompts, flashing lights, message push, etc.

[0097] The warning result is the final output after the warning process, which contains detailed information about the abnormal state of the motor, such as the time when the abnormality occurred, the type of abnormality, possible impact, etc.

[0098] Remote reporting is to use network communication technology to transmit the warning results from the local location of the motor to a remote server, monitoring center or other receiving end. In this way, even if the operator is not at the motor site, the abnormal information of the motor can be obtained in time to achieve remote monitoring and management.

[0099] After the state detection model analyzes the motor operating parameters, it obtains the state detection results. If the results show that the motor is operating abnormally, the system will start the warning program according to the pre-set warning rules. After generating the warning results, the system will use the network communication module to send the warning results to a remote target location. This target location can be the company's remote monitoring center or the maintenance personnel's mobile phone application. Through remote reporting, relevant personnel can be informed of the abnormal situation of the motor at the first time, and even if they are in a different place, they can respond in time, such as arranging maintenance personnel to go to the site to repair the motor.

[0100] It can be concluded from the above that this embodiment effectively avoids sudden downtime and ensures production continuity by real-time monitoring of motor status and timely warning of potential faults. The remote reporting function enables the operation and maintenance team to respond quickly. Regardless of the geographical location, they can obtain motor status information in the first place and take maintenance measures to ensure safe and stable operation of the motor, thereby improving operation and maintenance efficiency.

[0101] Corresponding to the motor operation energy efficiency detection method in the above embodiment, Figure 2 This is a structural block diagram of a motor operation energy efficiency detection device provided by an embodiment of the present disclosure. For ease of description, only the parts related to the embodiment of the present disclosure are shown. Figure 2 The motor operation energy efficiency detection device 20 includes: a sample determination module 21, a sample expansion module 22 and a detection model construction module 23.

[0102] Wherein, the sample determination module 21 is used to determine a plurality of normal samples and a plurality of abnormal samples of the motor operation parameters based on a clustering algorithm; A sample expansion module 22, configured to expand the plurality of abnormal samples according to the sample features of the plurality of abnormal samples to obtain a plurality of expanded abnormal samples; The detection model building module 23 is used to train the state detection model based on multiple normal samples and multiple expanded abnormal samples, and detect the motor running state to be detected according to the trained state detection model.

[0103] In one embodiment of the present disclosure, the motor operating parameters include multiple types of operating parameters; The sample determination module 21 is specifically used for: Perform clustering processing based on the parameter characteristics of each type of operating parameters to obtain clustering results corresponding to each type of operating parameters, and the clustering results include a normal sample set and an abnormal sample set; The normal sample sets corresponding to each type of operating parameters are collected to obtain multiple normal samples; The abnormal sample sets corresponding to each type of operating parameters are collected to obtain multiple abnormal samples.

[0104] In one embodiment of the present disclosure, the sample determination module 21 is further configured to: Determine multiple cluster centers; Determine the distance between each type of operating parameter and the operating parameters of multiple cluster centers based on the first distance measurement method to obtain multiple first distances; A clustering result corresponding to each type of operating parameter is determined based on the multiple first distances.

[0105] In one embodiment of the present disclosure, the sample expansion module 22 is specifically used to: Input multiple abnormal samples into the target generative adversarial network model to obtain multiple expanded abnormal samples; Among them, the target generative adversarial network model learns the sample features of multiple abnormal samples.

[0106] In one embodiment of the present disclosure, a generative adversarial network model includes a generator and a discriminator; The motor operation energy efficiency detection device 20 further includes: a sample expansion model building module; A sample expansion model building module is used to train the generator based on the historical abnormal sample set to obtain a trained generator; Inputting the random noise vector into the trained generator to obtain the first abnormal data set; Training the discriminator based on the historical abnormal sample set and the first abnormal data set to obtain a trained discriminator; Determine the target generative adversarial network model based on the trained generator and the trained discriminator.

[0107] In one embodiment of the present disclosure, the motor operation energy efficiency detection device 20 further includes: a parameter updating module; A parameter updating module, configured to determine a confidence level for the first abnormal data set based on the discriminator; If the confidence is greater than the first threshold, the parameters of the generator are retained; If the confidence is less than or equal to the first threshold, the parameters of the generator are updated.

[0108] In one embodiment of the present disclosure, the motor operation energy efficiency detection device 20 further includes: an early warning module; An early warning module is used to warn the motor operation status based on the status detection result to obtain the early warning result; Remotely report warning results.

[0109] See also Figure 3 , Figure 3 A schematic block diagram of an electronic device provided by an embodiment of the present disclosure. Figure 3The electronic device 300 in the embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303 and one or more memories 304. The processors 301, input devices 302, output devices 303 and memories 304 communicate with each other via a communication bus 305. The memory 304 is used to store computer programs, which include program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. The processor 301 is configured to call the program instructions to execute the functions of each module / unit in the above-mentioned device embodiments, such as Figure 2 The functions of modules 21 to 23 are shown.

[0110] It should be understood that in the embodiment of the present disclosure, the processor 301 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0111] The input device 302 may include a touch panel, a fingerprint collection sensor (for collecting the user's fingerprint information and fingerprint direction information), a microphone, etc., and the output device 303 may include a display (LCD, etc.), a speaker, etc.

[0112] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A portion of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store information about the device type.

[0113] In a specific implementation, the processor 301, input device 302, and output device 303 described in the embodiments of the present disclosure can execute the implementation methods described in the first and second embodiments of the motor operation energy efficiency detection method provided in the embodiments of the present disclosure, and can also execute the implementation methods of the electronic device described in the embodiments of the present disclosure, which will not be repeated here.

[0114] In another embodiment of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by the processor, all or part of the processes in the above-mentioned embodiment method are implemented, and the computer program can also be completed by instructing the relevant hardware through the computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0115] The computer-readable storage medium may be an internal storage unit of the electronic device of any of the aforementioned embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0116] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0117] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0118] In the several embodiments provided in the present application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces or units, or it can be an electrical, mechanical or other form of connection.

[0119] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present disclosure.

[0120] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0121] The above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present disclosure, and these modifications or replacements should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for detecting the energy efficiency of a motor, characterized in that: include: Determine multiple normal samples and multiple abnormal samples of motor operation parameters based on a clustering algorithm; Performing sample expansion on the multiple abnormal samples according to the sample features of the multiple abnormal samples to obtain multiple expanded abnormal samples; The state detection model is trained based on the multiple normal samples and the multiple expanded abnormal samples, and the motor operation energy efficiency is determined by detecting the motor operation state to be detected according to the trained state detection model.

2. The motor operation energy efficiency detection method according to claim 1, characterized in that: The motor operating parameters include multiple types of operating parameters; The method of determining a plurality of normal samples and a plurality of abnormal samples of motor operating parameters based on a clustering algorithm includes: Performing clustering processing based on parameter features of each type of operating parameters to obtain clustering results corresponding to each type of operating parameters, wherein the clustering results include a normal sample set and an abnormal sample set; Collecting the normal sample sets corresponding to each type of operating parameters to obtain the multiple normal samples; The abnormal sample sets corresponding to each type of operating parameters are collected to obtain the plurality of abnormal samples.

3. The motor operation energy efficiency detection method according to claim 2, characterized in that: The clustering process is performed based on the parameter characteristics of each type of operating parameters to obtain the clustering results corresponding to each type of operating parameters, including: Determine multiple cluster centers; Determine the distance between each type of operating parameter and the operating parameters of the multiple cluster centers based on a first distance measurement method to obtain multiple first distances; A clustering result corresponding to each type of operating parameter is determined based on the multiple first distances.

4. The motor operation energy efficiency detection method according to claim 1, characterized in that: The step of performing sample expansion on the multiple abnormal samples according to the sample features of the multiple abnormal samples to obtain multiple expanded abnormal samples includes: Inputting the multiple abnormal samples into a target generative adversarial network model to obtain the multiple expanded abnormal samples; Among them, the target generative adversarial network model learns the sample features of the multiple abnormal samples.

5. The motor operation energy efficiency detection method according to claim 4, characterized in that: The generative adversarial network model includes a generator and a discriminator; The motor operation energy efficiency detection method further includes: Training the generator based on a historical abnormal sample set to obtain a trained generator; Inputting the random noise vector into the trained generator to obtain the first abnormal data set; Training the discriminator based on the historical abnormal sample set and the first abnormal data set to obtain a trained discriminator; A target generative adversarial network model is determined based on the trained generator and the trained discriminator.

6. The motor operation energy efficiency detection method according to claim 5, characterized in that: Also includes: Determining a confidence level for the first anomaly data set based on the discriminator; If the confidence is greater than a first threshold, retaining the parameters of the generator; If the confidence level is less than or equal to a first threshold, the parameters of the generator are updated.

7. The motor operation energy efficiency detection method according to claim 1, characterized in that: Also includes: Based on the result of the state detection, the motor operation state is warned to obtain the warning result; Remotely report the warning results.

8. A motor operation energy efficiency detection device, characterized in that: include: A sample determination module, used for determining a plurality of normal samples and a plurality of abnormal samples of motor operation parameters based on a clustering algorithm; A sample expansion module, used to expand the plurality of abnormal samples according to the sample characteristics of the plurality of abnormal samples to obtain a plurality of expanded abnormal samples; The detection model building module is used to train the state detection model based on the multiple normal samples and the multiple expanded abnormal samples, and to detect the motor operation state to be detected according to the trained state detection model to determine the motor operation energy efficiency.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images